面向神经形态芯片的忆阻交叉阵列制备、脉冲控制与应用

随着全球数据量的激增和算法复杂度的提高，传统数字计算机的内存墙问题、功耗问题日益凸显，迫切需要研制出非冯·诺依曼架构的、高智能、低功耗、微型化的计算系统。本项目将融合材料科学、集成电路技术和深度学习智能算法的最新成果，开展忆阻神经形态芯片的关键问题研究：（1）采用先进的纳米级忆阻器件制备方法和集成工艺，研制高一致忆阻器件及交叉阵列，模仿大脑突触的功能；（2）融合两端/三端纳米忆阻器与CMOS集成电路工艺，构建高效率的新型存算一体架构；（3）结合CMOS工艺，制备新型神经形态电路芯片；（4）引入低功耗脉冲神经网络的优势，完成类脑的信息处理和编码；（5）结合深度学习算法，在芯片上实现深度神经网络，完成典型的图像处理任务。本项目研发出新一代神经形态系统，将有力推动存算一体高效计算架构的发展；研发的类脑芯片，将为新型的智能算法提供硬件支持。研究成果将为开发新一代人工智能计算系统提供重要的技术支撑。

With the increase of global data volume and algorithm complexity, the problem of memory walls and power consumption of traditional digital computers has become increasingly prominent, and there is an urgent need to develop Non-von Neumann architecture, high intelligence, low power consumption, and miniaturization computing system. This project will integrate the latest achievements of materials science, integrated circuit technology and deep learning intelligent algorithm to carry out the research on the key issues of memristor neuromorphic chips: (1) Adopting advanced nanometer memristor preparation methods and integration technology to develop high consistent memristive crossbar, imitating the function of brain synapse; (2) Fusing two-terminal / three-terminal memristor and CMOS integrated circuit technology to build a high-efficiency novel Computation-in-Memory architecture; (3) Combining CMOS technology to prepare a new neuromorphic chip; (4) Introducing the advantages of low-power-consumption Spike Neural Networks to complete the brain-inspired information processing and coding; (5) Combining deep learning algorithm to implement deep neural networks on the chips to complete typical image processing tasks. The project can help to develop a new generation of neuromorphic system, which will strongly promote the development of Computation-in-Memory architecture. The brain-inspired chips will provide hardware support for new intelligent algorithms. The research results will provide important technical support for the development of a new generation of Artificial Intelligence Computing System in China.